Article

Urban Studies 2019, Vol. 56(8) 1681–1700 Ó Urban Studies Journal Limited 2018 Protesting iconic megaprojects. A Article reuse guidelines: sagepub.com/journals-permissions discourse network analysis of the DOI: 10.1177/0042098018775903 evolution of the conflict over journals.sagepub.com/home/usj 21

Melanie Nagel University of Constance,

Keiichi Satoh University of Constance, Germany

Abstract The past decades witnessed enormous transformations in the built environment of cities, and one of these trends is the development of iconic megaprojects. Public protests against these proj- ects occur frequently, and scholars in urban governance have diagnosed this as an emerging ‘post- political’ condition, that is, as a sign of a deficient democratic politics. Others criticise this kind of reasoning as a ‘post-political-trap’ (Beveridge and Koch, 2017), and demand more research. This article responds to this debate with an empirical study of the popular protests against the infra- structural public transport project in Germany. We apply discourse network analysis to investigate the evolution of the discourse, illuminate multiple dynamic connections between issues and actors, and apply factor analysis to identify the key issues of the conflict. Our study complicates and qualifies the arguments for a ‘post-political’ state of urban politics.

Keywords built environment, discourse network analysis, infrastructural projects, networks, planning, policy, transport

᪈㾱 䗷৫ࠐॱᒤˈ෾ᐲᔪㆁ⧟ຳਁ⭏Ҷᐘབྷਈॆˈަѝањ䎻࣯ᱟḷᘇᙗབྷර亩ⴞⲴᔰਁDŽ䪸 ሩ䘉Ӌ亩ⴞⲴޜᔰᣇ䇞⍫ࣘ仁㑱ਁ⭏ˈ෾ᐲ⋫⨶ᆖ㘵ሶ↔㿶Ѫа⿽↓൘ࠪ⧠Ⲵ “ਾ᭯⋫”⣦ ߥˈণ≁ѫ᭯⋫н䏣ⲴḷᘇDŽަԆӪࡉᢩ䇴䘉⿽᧘⨶ᱟ “ਾ᭯⋫䲧䱡 ”˄Beveridge઼Kochˈ 2017˅ˈᒦф䴰㾱ᴤཊⲴ⹄ウDŽᵜ᮷䙊䗷ሩᗧഭޜޡӔ䙊ส⹰䇮ᯭ亩ⴞĀᯟമ࣐⢩ 21āⲴ 㗔Շᣇ䇞⍫ࣘ䘋㹼ᇎ䇱⹄ウˈഎᓄҶ䘉൪ҹ䇪DŽᡁԜ䘀⭘Ā䈍䈝㖁㔌࠶᷀āᶕ⹄ウ䈍䈝Ⲵ ╄ਈˈ䱀᰾䰞仈о㹼ࣘ㘵ѻ䰤Ⲵཊ䟽ࣘᘱ㚄㌫ˈᒦᓄ⭘ഐ㍐࠶᷀ᶕ⺞ᇊߢケⲴޣ䭞䰞仈DŽ ᡁԜⲴ⹄ウሶޣҾ෾ᐲ᭯⋫ “ਾ᭯⋫”⣦ᘱⲴҹ䇪᧘ੁᴤѪ␡ޕ༽ᵲǃᕅޕᴤཊ䲀ࡦᶑԦⲴ 䱦⇥DŽ ޣ䭞䇽 ᔪㆁ⧟ຳǃ䈍䈝㖁㔌࠶᷀ǃส⹰䇮ᯭ亩ⴞǃ㖁㔌ǃ㿴ࡂˈ᭯ㆆˈӔ䙊

Konstanzer Online-Publikations-System (KOPS) URL: http://nbn-resolving.de/urn:nbn:de:bsz:352-2-t4wh6d0ob1df5 1682 Urban Studies 56(8)

Received January 2016; accepted April 2018

Introduction tracks when completed will serve high speed regional and European express and local Research on urban politics and urban gov- commuter trains, turning Stuttgart into an ernance has grown over the last three important transport hub situated in the heart decades (MacLeod, 2011), mirroring the of Europe. The name for the proposed proj- immense transformations undergone by cit- ect, Stuttgart 21 (S21), soon became a syno- ies themselves. One of these trends is ‘a pro- nym for the German word ‘Wutbu¨rger’, liferation of megaprojects of iconic which means ‘furious citizens’. The infra- development and associated infrastructures’ structural project was discussed emotionally, (MacLeod, 2011: 2630; see also Douglass protests were staged, and things reached a and Huang, 2007; Graham and Marvin, peak when the protests escalated violently 2001; Young and Keil, 2010). Over the between protestors and the police on 30 course of the last two decades protests and September 2010. While this appears in line discontent in the context of the planning with the ‘democratic deficits’ view, a referen- and building of mega urban infrastructure dum held later in November 2011 showed projects have become a common occurrence, that, in fact, most citizens were pro S21. The and scholars attribute such public resistance incongruity between the vehemence of the to democratic deficits and limited opportuni- protests and majority support (as demon- ties for public participation (Marres, 2005, strated by the referendum results) bears cited in Legacy, 2016). Most strategic urban explaining. This paper applies insights drawn megaprojects currently under construction from theories of discourse coalitions to an are, according to Moulaert et al. (2001), empirical study that combines discourse dominated by ‘neoliberal’ efforts by ‘elite- analysis and network analysis. By analysing formations’ to build ‘the entrepreneurial major newspaper accounts of diverse groups city’. Other scholars also criticise the lack of of actors (including organisations, social democratic consultation and the absence of movements, civil society groups, and politi- coherent social policy to guide new urban cal parties) over a ten-year period, and by development politics, as well as the lack of using factor analysis, our empirical study integration of megaprojects in urban pro- shows which actors were involved at what cesses and planning systems (cf. Moulaert time, and which issues were important to et al., 2003; Salet, 2008; Swyngedouw et al., them. 2002; cf. Graham and Marvin, 2003). Thus, the study contributes to studies on In this article, we challenge the prevailing urban governance, one of the key concepts of view of the post-democracy argument that urban studies (McCann, 2017). The key ques- democratic deficits alone are to blame for tion of governance is not only why cities are strong discontent and polarisation. We ana- the way they are now, but also ‘[.] how they lyse the conflict surrounding the planned are made to be the way they are, through the construction of an underground railway sta- concerted actions of the state, other public tion in the city of Stuttgart, Germany, whose and private institutions, social movements,

Corresponding author: Melanie Nagel, Department of Public Policy and Administration, University of Constance, Fach D81, Konstanz 72160, Germany. Email: [email protected] Nagel and Satoh 1683 civil society and the practices of everyday life’ However, our study of S21 shows that (McCann, 2017: 313). The governance there is no straightforward link between citi- approach focuses on the ability of political zen protest and democratic deficits, and that, actors to act, and therefore it is closely related instead of a tendency toward elite-formation to policy output (cf. Gissendanner, 2004: 47). as scholars have warned, popular and elite The case of S21 belies the claims of recent lit- preferences were in fact aligned. Thus, it erature on the post-political condition affirms some of the criticisms made of the (Mouffe, 2005; Rancie` re, 2001), especially in ‘post-political-trap’ (Beveridge and Koch, urban studies, which argues that the new dis- 2017). course platforms created alongside the state This paper is structured as follows. In the are signs of shifted politics, that is, a resis- first section, an overview of the history of tance to the state’s inability to take popular S21, the main actors and their particular preferences into account (Blu¨hdorn, 2013; positions, and a review of scholarly Boland, 2014; Bylund 2012; Oosterlynck and approaches will be given. The next part Swyngedouw, 2010). The post-political fram- introduces the general theoretical approach ing of urban infrastructures introduces of discourse network analysis. Then, we according to Legacy (2016) a strong contrast explain the different steps of the data collec- between state planning and citizens’ counter- tion process, selection of the data sources, action, and other scholars argue that the lack coding procedure, and introduce the method of transparency and lack of democratic con- of the discourse network analysis, before trol in the development of iconic megaprojects presenting the findings. First, we explain ‘play a crucial role in the erosion of democ- who the important actors and discourse racy’ (Vento, 2017: 68). The expressions ‘post- coalitions are and point out their positions democracy’ (Crouch, 2004), ‘post-politics’ visualised in an actor-by-issue network. In (Mouffe, 2005) or ‘post-political’ (Rancie` re, the second part, we present the factor analy- 2009) illustrate ‘[.] a contemporary demo- sis of issue-by-issue networks and the entan- cratic condition in which genuine contestation glement of issues connected to these actors. and conflicting claims [.] are not apparent’ Lastly, we recap our findings and conclude (Beveridge and Koch, 2017: 32–33). with a discussion of how these findings

Figure 1. Timeline of the key events of the project Stuttgart 21. Source: Nagel (2016: 156; graphics modified). 1684 Urban Studies 56(8) contribute to urban studies literature by environmental organisation NABU (acronym challenging the argument of post-democracy for Naturschutzbund Deutschland,‘Nature and post-politics. Protection Alliance Germany’) and the The time frame of this study ranges from Mineralbad Berg (Berg mineral spa, a public 2002 until 2011, for two reasons: analysing swimming pool) went to court against the the policy process by means of a policy ana- project in 2005, without success. In 2006, lysis requires a time span of at least 10 years negotiations between DB, the state govern- (e.g. Jenkins-Smith et al., 2014) and in our ment of Baden-Wu¨rttemberg and the city of case study, different crucial phases, as Stuttgart were held, followed by a funding explained in the subsequent part, could be agreement in 2007, which was passed with a identified during that time. Figure 1 shows a total budget of e4.8 billion. In 2006, 67,000 timeline of the major events. people signed a petition against this project, and weekly demonstrations in Stuttgart drew many thousands of protestors. Brief history of Stuttgart 21 The situation escalated in October 2010: S21 is an ambitious infrastructural project thousands were protesting in the Palace by the biggest railway company in Germany, Park, when the conflict between protestors (DB), the city of Stuttgart, and police escalated. Images of the police the state Baden-Wu¨rttemberg and the using water cannons against protestors, German Federal Republic. The southern including elderly people and children, were German city of Stuttgart is situated near the broadcast on media nationwide. This shock- geographical centre of Europe and the ing event gained nationwide attention, and planned railway connection is part of the led to a mediation process, which lasted Magistrale for Europe – a high-speed railway from 22 October to 30 November 2010. track between , Strasbourg, Stuttgart, In March 2011, a new Baden- , Munich, Vienna and Bratislava. A Wu¨rttemberg Parliament, consisting of a rapid transit route from Wendlingen to Ulm, coalition of the Green Party and the Social two other cities in Germany, is also included Democrats, was elected, toppling the incum- in this project. bent CDU (Christlich Demokratische Partei In the 1990s, big urban infrastructure proj- or Christian Democratic Party). The ects were planned in almost every metropoli- Fukushima nuclear catastrophe and discon- tan region in Europe, and what Banister and tent with the S21 project were among the Hall (1993) call the ‘second railway age’, with reasons why voters decided to drop the for- inner-city railway stations surrounded by mer conservative government after 58 years other buildings and infrastructure, became the of dominance. A few months later, the newly centre of ambitious urban redevelopment elected government fulfilled its campaign plans. The planning of S21 started in the promise to hold a referendum on S21. In 1990s with a proposal by the traffic expert November 2011, in the state of Baden- Gerhard Heimerl. In 1994, the project S21 Wu¨rttemberg, a majority of citizens surpris- was presented to the public and in 1997, the ingly voted in favour of S21. master plan for the city of Stuttgart was devel- To put it in a nutshell, S21 is not only a oped. The infrastructural project was the main modernisation of the old railway station, it is issue of the mayoral (Oberbu¨rgermeister) elec- an enormous financial investment, passion- tions in Stuttgart in 2003 and in 2004, because ately opposed by environmental activists and of cost explosions forecasted by various civil society groups, but, surprisingly, experts, which officials denied. The favoured by many citizens. From this brief Nagel and Satoh 1685 history, several questions arise: Why did this enables us to see changes to discourses over project become the subject of such heated time, and offers a more nuanced overview of debate? How can we explain the evolution of actors and issues. The combined analysis of this emotional and polarised process? Which actors as well as the different issues that con- lessons can be learned for other megaprojects nect them allows us to get to the heart of the and the diagnosis of post-democracy? conflict, and to address several dimensions Urban transport planning is ‘inherently of urban conflict in a single analysis. Third, political’, with priority settings and decisions ideas and beliefs play an important role in on huge investments that serve some groups policy-making (Bradford, 2016: 659), and better than others (Legacy, 2016: 3108). discourse network analysis contributes to Studies of S21 scrutinise the emphasis on our understanding of the role they play. justice sensitivity (Rothmund et al., 2014), This perspective (and its associated metho- the use of Twitter for the organisation of dology) is still underdeveloped in urban poli- collective action (Jungherr and Ju¨rgens, tics. Finally, we apply a new and innovative 2014), Facebook activities of proponents approach of discourse network analysis to a and opponents (Maier, 2013), voting beha- complex issue in urban politics, which allows viours and the NIMBY effect (cf. Vatter and us to trace the main lines of contention and Heidelberger, 2013). The most comprehen- dissent, without oversimplifying. sive discussion of the conflict over Stuttgart 21 and the inherent fundamental disagree- Discourse coalitions: Theory and ment over future urban planning and trans- portation policy, costs and benefits of these methods expensive megaprojects and the procedures To understand urban governance in the con- of democratic decision-making is by Novy text of thorny problems and complex policy and Peters (2013). Using a similar case of initiatives, Neil Bradford (2016) argues that public protests against an airport expansion, under conditions of uncertainty, ideas play Eike-Christian Hornig (2017) refers to the an important role ‘in motivating collective Political Opportunity Structure Approach action, channelling policy resources, and and offers a new model called Vertically shaping prospects for collaboration’ Asymmetric Policies (VAPs) to explain the (Bradford, 2016: 660). He proposes an ana- ‘democratic dilemma of multi-level politics lytical framework called discursive localism between system effectiveness and citizen par- for a better understanding of politics and ticipation’ (Hornig, 2017: 325). Our paper policy in cities. That is, when issues are builds on these studies, and likewise asks highly complex and uncertain, and the time questions concerning democracy and partici- window is especially long, government con- pation. However, our study carries out a stellations alternate and change (Bradford, more in-depth analysis of the evolution of 2016: 659), not just in relation to these long- the discourse to disclose the underlying term issues, but also as short-term strategic beliefs and ideas driving the public and visi- considerations, such as before elections. ble actions and statements. The ‘argumentative turn’ in political and We therefore focus on a multidimensional social sciences has shown that ideas and perspective, which carries out a dynamical beliefs play an important role for collective analysis of actors, their issues of concern, action and the channelling of policy and the connections amongst issues and resources. Different groups of actors are actors over a period of ten years. The evolu- involved in the decision process, and their tion of the conflict over a long period unique underlying ideas and perceptions take 1686 Urban Studies 56(8) discursive shape. According to discourse hailing from civic and non-governmental analysis (Hajer, 1993, 1995, 2002) and discur- organisations and interest groups (Kenis sive institutionalism (Schmidt, 2008; Schmidt and Schneider, 1991). and Radaelli, 2004), the political process can be seen as a struggle over discursive hege- mony between conflicting discourse coali- Method tions (Leifeld and Haunss, 2012). The groups To measure and visualise the evolution of a of actors that share the same framework of particular political discourse, such as S21, interpretation and ideas are called discourse researchers have employed a tool called dis- coalitions (e.g. Bulkeley, 2000; Fischer and course network analysis (see Janning et al., Forester, 1993; Hajer, 1993, 1996, 2002; 2009; Leifeld, 2009, 2010, 2012, 2013). Szarka, 2004), and within any society, several Discourse network analysis is a dynamical discourse coalitions always struggle for dis- and longitudinal analysis of political dis- cursive hegemony (Muller, 2015). course (Leifeld, 2010: 13). It is an approach The discourse theoretical approaches are that combines qualitative content analysis derived mainly from Michel Foucault’s work and quantitative social network analysis, on post-structuralism and Ju¨rgen Habermas’ and at the same time offers a new possibility communicative discourse ethics. ‘Discourse for conceptualising and measuring the ‘co- theory starts from the assumption that all evolution of actors and concepts in a actions, objects, and practices are socially dynamic way’ (Leifeld, 2010: 4). Social net- meaningful and that these meanings are work analysis serves as an analytical toolbox shaped by the social and political struggles for relational analysis (Kenis and Schneider, in specific historical periods’ (Fischer, 2003: 1991), and policy debates can be operationa- 73). This article uses Maarten Hajer’s lised by aligning the actors by their common Discourse Coalitions approach (1995, 2002), claims (Koopmans and Statham, 1999). which advocates investigating language and Furthermore, both approaches integrate text analysis, as well as the analysis of the actors and concepts (Leifeld and Haunss, constellations of actors. If a discourse coali- 2012: 389). tion wants to be successful, it has to domi- Graph theory and matrix algebra serve as nate the public discourse, and its dominance a foundation for many conceptualisations of will be reflected by institutional practices in social network analyses (Hage and Haray, the political domain (Hajer, 1993: 48). The 1983; Iacobucci, 1994). Actors and relations dominant coalition is prominent in the between them are modelled by graphs con- media, and is therefore able to integrate its sisting of nodes, which are connected by core frames into a consistent storyline, which edges. In other words, the nodes represent is superior to the storyline of its opponents. actors (e.g. organisations, persons, compa- This paper maps discourse coalitions by nies), and the edges represent relations (e.g. means of discourse network analysis, which exchange of knowledge, exchange of money, combines discourse analysis and policy net- sharing the same beliefs). work analysis. By policy network we mean As described in detail by Leifeld and all actors who participate in the public pol- Haunss (2012: 390ff) and Leifeld (2013: icy process, as well as the relations amongst 174), a discourse network has the basic form these actors. These are not only actors of the of an affiliation network. The visualisation government and parliament, but also, of the network is a graph with actors as ver- because of resource dependence and interest tices and the number of shared issue posi- interrelations, a variety of heterogenic actors tions as edge weights between them (Fisher Nagel and Satoh 1687 et al., 2013: 530). The basic matrix is an 1999; Rucht et al., 1999). Nevertheless, it actor-by-concept affiliation matrix, which is should not be forgotten that newspaper con- converted into a square matrix (co-occur- tent is not created for the purpose of social rence matrix) where all actors (rows and col- scientific research (see Ortiz et al., 2005: 397). umns in the matrix) are connected by the Questions of reliability and validity have to number of their shared concepts. In the cells be taken into account, and the advantages of the matrix, 0 stands for no shared com- and disadvantages of newspaper data have to mon concept, numbers (1 to n) in the cells be considered (Baumgarten and Grauel, for shared common concepts and the diago- 2009). Factors including the intensity of an nal entries are left blank. There are shared event, conflicting elements, and the signifi- common concepts if both agree or disagree cance of actors involved could increase media on one or more concepts (categories of state- attention. Media structure and media outlets ments). These co-occurrence networks are could also influence individual newspaper interpreted as undirected and weighted issues, as might be found in the case of atten- according to the number of shared concepts. tion cycles, which refer to the phenomenon The visualisation with network diagrams of whereby certain issues gain prominence for a discourse networks is thus a good opportu- period of time before fading into obscurity nity to operationalise policy discourses. (McCarthy et al., 1996; McCombs and Shaw, 1972). Newspaper data are but one discursive arena among many (Hilgartner and Bosk, Data collection 1988), and the question is if this arena is rep- The data collection consists of newspaper resentative of all the others (Baumgarten and articles on ‘Stuttgart 21’ published in four Grauel, 2009). For a dominant issue, such as different German newspapers from 2002 the case of S21, the number of articles should through 2011.1 The software program be sufficient to comprehensively investigate Discourse Network Analyzer2 (or DNA) was the public media discourse. The descriptions used for organising and coding the data. This of the selection methods in the following sec- computer program allows the quantitative tions consider these limiting factors, the pos- coding of actors and their statements, and sible effects of biases in our study, and offers the possibility of exporting network possible limitations of our findings. data for the visualisation of the relationships between actors in a discourse field, and how Selected time frame and newspaper they are connected by shared arguments. For the discourse network analysis, we chose dif- sources ferent steps for collecting the data. The key To obtain sufficient data on phases of politi- issues are the selection of the time frame, the cal emergence and maturing of the issue, newspaper source, and the coding procedure, escalations during the protests, governmental which will be explained below. changes, and public participation (as Collecting data from newspaper media is explained earlier in the history of S21), we a popular and easily available source for selected a time span from 2002 until 2011. analysing social events (Earl et al., 2004). The selection of the newspapers was deter- This simple access to big data sources, with mined by the following considerations. First, daily up-to-date availability, offers the possi- whether they were available in digital full-text bility of analysing dynamical data over many archives; second, if they corresponded to the years. Many scholars use newspaper data ‘quality press criterion’ (Barranco and (e.g. Franzosi, 1987; Koopmans and Rucht, Wisler, 1999; Koopmans, 1996; Kriesi, 1995; 1688 Urban Studies 56(8)

Leifeld, 2013), that is, they must fulfil condi- the data through alternative and indepen- tions of a wide publication range, political dent measures. One scientist coded all the moderation, and a good reputation. The data data in this study, which might pose a prob- collection consists of newspaper articles on lem of reliability rather than of validity. Our ‘Stuttgart 21’ in four different German news- attempt to reduce this bias was to have papers: two regional newspapers (Stuttgarter another scientist randomly control the data, Nachrichten and Stuttgarter Zeitung) and two and in case of deviation, a solution was dis- national newspapers (Frankfurter Allgemeine cussed and noted in the codebook. Zeitung and Su¨ddeutsche Zeitung). Overall, 556 articles, with 2077 statements All articles were selected if they contained were coded (see Figure 2 coded categories and the search term ‘Stuttgart 21’ in the title or number of statements). These statements were the first lines, and if they were relevant for the coded into 155 categories, which were divided case. This means that all articles dealing with into 34 meta-categories for factor analysis (see the issue S21 were selected, except for articles meta-categories in Table 1; a detailed over- describing, for example, a theatre play dealing view of all categories and meta-categories is in with S21. In total 556 articles dealing with the the Appendix, available online). case of S21 were selected, each of which com- prises more than 500 words. Research findings This article explores how the evolution of The coding procedure the polarised debate on S21 can be under- For the coding procedure four kinds of stood, also, in the light of a post-democracy information were coded: the statement, the debate. As noted above, the complex case of name of the actor, the organisation the actor S21 was coded into 155 different categories belongs to, the category the statement of arguments, divided into 34 different meta- belongs to, and a dichotomous variable categories. Figure 2 shows the coded cate- ‘agreement’ or ‘disagreement’. The categori- gories and number of statements. sation of the statements was coded manually using predefined codes (deductive coding) and expanded during the coding procedure Actors and discourse coalitions (inductive coding) (for more details of the This study identified 319 persons belonging coding process, see Nagel, 2016). to 91 organisational actors engaged in policy Researcher-bias through coding and data entry errors is a critical point, because the coding procedure requires an accurate and intensive coordination as well as theoretical and conceptual clarity. The question of whether content analysis data is reliable has to be answered by content analysis metho- dologists at all times. The underlying prob- lem is called intercoder reliability (Franzosi, 2004: 182). Random errors (reliability) from newspaper data and non-random errors (validity) occur and have to be limited and, Figure 2. Coded categories and number of if possible, eliminated. Franzosi (2004: 183) statements. recommends focusing on the validation of Source: Author’s own graphics. ae n Satoh and Nagel Table 1. Top five categories in each year (%).

Rank 1 2 3 4 5

2002 The building of the Not enough profit The building of the S21 will/should be expected with S21 S21 will/should be delayed (60.0) (no) (20.0) delayed (no) (20.0) 2003 Keep the The building of the The building of the not enough profit K21 alternative construction plan S21 will/should be S21 will/should be expected with S21 proposal (no) (8.3) (50.0) delayed (16.7) delayed (no) (8.3) (8.3) 2004 More costs will be More costs will be More connected cities not enough profit Keep the construction expected with S21 expected with S21 and Europe with S21 expected with S21 plan (no) (7.5) (no) (15.1) (15.1) (9.4) (no) (7.5) 2005 Keep the More costs will be More costs will be EU should take more Keep the construction construction plan expected with S21 expected with S21 cost (9.8) plan (no) (6.1) (14.6) (no) (11.0) (9.8) 2007 Keep the not enough profit More connected cities More costs will be not enough profit Federal State construction plan expected with S21 and Europe with S21 expected with S21 expected with S21 should finance (16.6) (no) (10.6) (10.2) (6.8) (6.8) more (6.8) 2008 not enough profit Referendum should be City should finance Keep the construction More costs will be expected with S21 done (6.7) more (no) (6.3) plan (5.7) expected with S21 (no) (7.3) (5.1) 2009 Keep the More costs will be More costs will be Public communication Opponents successful construction plan expected with S21 expected with S21 should be improved and enough reason (10.8) (no) (9.5) (8.9) (8.2) (no) (7.0) 2010 Keep the Keep the construction Arbitration should be Referendum should be Social peace is construction plan plan (12.2) done (7.9) done (7.1) important (4.5) (no) (13.5) 2011 Referendum should result of stress test is Keep the construction Keep the construction result of stress test is be done(12.8) valid (no) (12.6) plan (10.5) plan (no) (8.7) valid (8.5) Total Keep the Keep the construction Referendum should be not enough profit More costs will be construction plan plan (no) (8.5) done (6.4) expected with S21 expected with S21 (11.3) (no) (5.5) (5.2) 1689 1690 Urban Studies 56(8) discourse between 2002 and 2011. These instead of 16 platforms. Klaus Arnoldi, a actors were included when they made at politician and S21 opponent, stated: ‘Instead least one statement in one or more of the of a 33 km tunnel, as seen in Stuttgart 21, 556 newspaper articles concerning S21. We only a ten km tunnel-route was planned, argue that all actors appearing in the media which would be cheaper and associated with are important for the issue of S21, since they lower cost risks’ (statement of Klaus will warrant mention in the media if they are Arnoldi, in Stuttgarter Zeitung, 30 January important to the case. These actors belong 2007; all statements translated by authors). to various political parties, organisations, The following statements illustrate the social movements, associations, or DB. opponents’ views: ‘The only problem is that the reconstruction simply cannot be financed’ Central actors and their positions (statement of , politician, in Stuttgarter Zeitung, 26 October 2006). ‘One There are two central groups of actors: cannot cut back on regional trains and simul- opponents and proponents of the project taneously strive for a project devouring bil- (see Appendix, available online, for the lions’, argued Gerhard Pfeifer, managing entire list of the actors involved in this ana- director of the BUND-regional-union. Or, as lysis). Table 2 shows the most important Boris Palmer put it: ‘My great concern is that groups or organisations and their respective thePrimeMinister,aswellasthemayorof positions. The importance and relevance of Stuttgart, , want to enforce these actors is assessed on the basis of their Stuttgart 21 by all means, just for the sake of frequent occurrence in the newspaper arti- thereputationandtheprestigeofStuttgart’ cles analysed. All underlying information (in Stuttgarter Zeitung, 26 October 2006). comes from newspaper articles and website In October 2007, a broad range of citizen research. There are actors from both sides, and environmental organisations, supporters pro and contra S21: political parties, social of the Green Party and grassroots initiatives movements and civil society groups, associa- such as ‘Living in Stuttgart’ organised a peti- tions, and officials of the DB. tion and public demonstrations. This petition was signed by 67,000 people, and the political Opponents of Stuttgart 21. The opponents of struggle over whether the matter could be S21 are a heterogeneous group of actors, decided by a local referendum began. Legal mostly environmental activists, traffic clubs, experts denied the validity of the referendum, left-wing and green political parties and civil arguing that the project was not only financed society groups, all allied in their opposition by the city of Stuttgart but that it was not for to the infrastructural project. Opponents of the city to make the final decision. S21 demanded a different model called When the protests against the project ‘Railhead 21’ (Kopfbahnhof 21). This alterna- escalated in 2010, a mediation procedure tive model proposed a renovation of the cur- gave both sides the opportunity to submit rent railway station and the building of a few their arguments in a moderated and publicly new railway tracks. According to the S21 broadcast process. In March 2011, the elec- opponents, the costs would be less expensive tions, strongly influenced by the controver- than the underground construction, and the sial issue and the Fukushima nuclear performance could possibly be better than accident in Japan, swept the Green Party the planned S21 station. Additionally, it and Social Democrats into power, and would allow railway tracks to travel through brought about institutional change and a the station, but would contain only eight supposed harmonisation in the polarised Table 2. Central organisational actors and their positions on S21. Satoh and Nagel

Organisation information/ name Description Position

Parties CDU Christian Democratic Union Conservative party, proponent of the project Pro S21 In government for more than 50 years SPD Social Democratic Party Half of the party is for and the other half is against the project Pro/con S21 S21 In government together with the Greens since 2011 The Greens Environmental Party Against the project S21 Con S21 Governing with the SPD since 2011, Prime Minister of their party FDP Liberal Democratic Party Proponent of the project S21 Pro S21 Governing together with the CDU until 2011 SO¨ S Stuttgart Ecological and Social Party in the city parliament of Stuttgart, against the project S 21 Con S21 Social movements and civil society groups Action Alliance Numerous members, e.g. civil society Action Alliance against S21 organised protest and collected Con S21 against S21 group Living in Stuttgart, BUND signatures for a petition on a referendum (Union for environment and nature Petition with the demand of a referendum was rejected due to protection), /The Green legal reasons Party, Pro Railway Regional Union Stuttgart, Traffic Club Germany BW, Community SO¨ S and The Left Party in the city parliament of Stuttgart . Park Protectors www.parkschuetzer.de Very active and partly radical protestors with the goal to protect Con S21 the trees in the Palace Park. Tree protest levels, demonstrations, sitting protests and people chaining themselves to the trees BUND Environmental Protection Organisation Against S21 because of suspected environmental damage Con S21 Germany Association Verein Pro Stuttgart 21 Internet webpage and Facebook page Pro S21 Pro Stuttgart 21 Action Alliance Aktionsgemeinschaft Stuttgart 21 Intended purpose is newspaper advertising Pro S21 pro S21 Respect 21.de www.respect21.de Criticises the conflict and claims respect of both sides neutral Associations Change Stuttgart Umkehr Stuttgart Alliance of: NABU, VCD, BUND – criticises ecological damage of Con S21

the project 1691 (continued) 1692 Urban Studies 56(8)

discourse (Nagel, 2016). In November 2011, a state-wide referendum was held. The out- come was that the required quorum for halt-

Con S21 Pro S21 ing the project was not achieved – which means most of the citizens, surprisingly, voted in favour of S21.

The proponents of Stuttgart 21. The actors sup- porting S21 consist mainly of the conserva- tive political parties CDU and FDP (Freie Demokratische Partei or Liberal Democratic Party), the Chamber of Industry and Commerce, the Association Region Stuttgart, the Association Pro Stuttgart 21, and the Action Alliance Pro S21. They back the plan for a new railway station, the under- ground construction of tracks, and the infra- structural project in general. They are especially interested in new possibilities for urban development and the building of a valuable new inner-city quarter, made possi- ble by the demolition and underground replacement of tracks. They reject the alter- native model Railhead 21 because, as Ulrich advantages, proposes the alternative concept21) K21 (Kopfbahnhof S21 Defends the advantages of the railway projectMu¨ller, Pro S21 Minister of Transport for Baden- Wu¨rttemberg, explained, ‘What critics mis- judge: Incredibly large investments would be needed to preserve the status quo, for exam- ple for the modernization of railway control centres. To our knowledge approximately 800 Million e would be needed – without the benefits of Stuttgart 21’ (in Stuttgarter sler Experts from both sides Pro/con S21 ¨ Zeitung Ro , 18 February 2004). Sebastian Sage,

+ architect and S21 supporter, was of the opin- ion that ‘If Stuttgart 21 was rejected, another ten years without any improvements around the railway node in Stuttgart would be lost’ Verband Region Stuttgart(Deutsche Bahn) Public corporation with voted members, sees many advantages of (in Stuttgarter Zeitung, 5 October 2006). To sum up, by unpacking the content of the discourse over Stuttgart 21, we could find out more about different views advanced by opponents and proponents of

Continued S21, and the different priorities and visions that guide their arguments. Opponents feared cost explosions, environmental dam- Others The ChurchThe airport of StuttgartExternal experts Die Kirche e.g. Vieregg Neutral position Connection of the new tracks for a better junction Pro S21 neutral Table 2. OrganisationVCD information/ name Traffic Club Germany Description Argues against S21, criticises the loss of tracks and lack of Position Association Region Stuttgart IHKRailway Organisation DB Chamber of industry and commerceage German Railway Organisation and States that S21 is the most importantloss infrastructure project of control Pro S21 over decisions on Nagel and Satoh 1693 urban planning. They were frustrated with discourse network data and explores the evo- the current political establishment and called lution of discourse coalitions over time. for more public participation. Proponents were annoyed by the protestors, and feared Evolution of the discourse the state parliament would lose control over the decision process, as well as the stagna- To trace the evolution of perspectives on the tion of future infrastructural investments. controversial issue, we now focus on the Our empirical findings show that the funda- above-mentioned meta-categories of the mental disagreement is only partly connected discourse. to the wish for more public participation, and Looking at the number of statements in other issues play a crucial role.3 This empirical each year, we recognise the following consid- finding is in contrast to the commonly used erations. First, the number of statements explanation in urban studies literature, that is, increased in 2010, which suggests that S21 a lack of transparency and democratic control became an increasingly important issue in (Vento, 2017: 68). public discourse. Second, when looking at A further question remains, namely, the trend, two peaks can be found – the first of how these coalitions and issues have devel- in 2007 and the second in 2010. Now, to oped over time. The next section analyses the have a closer look at the discourse coalitions

Figure 3. Actors and issues more or equal to 5. Source: Author’s own visualisation with Visone. 1694 Urban Studies 56(8) and their relevant issues, they will be illu- construction became highly polarised, espe- strated in the following discourse networks cially in 2009, when the top issue was the (see Figure 3). retention of the construction plan, but the second and third issues concerned disputes over the expected costs. Next, in 2010, the Discourse actors and relevant issues discussion about how to reach a consensus By visualising the actor-by-category net- given the polarised views (‘arbitration work, the discourse network approach should take place’) became more common. allows us to analyse actors and categories in Finally, in 2011, the issue ‘referendum one graph. The round nodes represent the should be done’ was the third most actors and the quadrangular nodes symbo- common. lise the issues; the size of the nodes shows their respective importance. The width of Factor analysis of categories and actors the lines shows the weight of their relations Factor analysis identifies the most impor- – thin lines mean that there are only five tant issues for the actors, the crucial common issues or actors, whereas thick con- dimensions of the conflict. In the previous nections represent strong relations with section, we have seen that the various state- many shared issues. We have to note that ments (including the execution of the plan, 4 the Visone-algorithm places the nodes next cost of the construction, and idea of a to similar nodes, and with its quick-layout referendum) were interconnected, and have 5 function produces a reasonable layout. ebbed and flowed in importance during the According to the similarity and size of the period under investigation. Now, the ques- nodes, we can interpret the visualisation tion remains of how these diverse state- below as follows. The central actors are the ments are connected to one other. How can CDU and the DB with their common issue we get the macro view on how the discus- ‘Keep the construction plan’, and the SPD sions evolved? and the Green Party with their shared issue To answer the question, we calculated a ‘Referendum should be done’. factor analysis. In our data, each organisa- This visualisation reveals the connections tion’s statement is coded with 1 if the organi- between actors and issues, but on closer sation agrees with the statement, and 21if inspection, we have to analyse how the con- they disagree with it. We aggregated the tents of the discussions developed over time. categories by organisation, which means by To do so, we will more closely examine the the direction in which the relevant organisa- top five categories of issues each year in the tion is headed. Based on this aggregated following section. data, we conducted a factor analysis, with a When looking at the top five categories minimum residual solution with an oblimin each year, we discover the following. First, rotation, using the R package psych v. 1.6.12 in 2002, objections against the construction (Revelle, 2016). were stated, and since 2004, the costs of the The following two main factors were iden- construction have become the main issue of tified (Table 3). Factor 1 consists of state- the dispute. Furthermore, in the first peak in ments such as ‘Keep the construction plan’, 2007, the financial issue including the costs ‘More connected cities and Europe with and profits of the construction was the issue S21’, and ‘Public communication should be mentioned the most. From 2008 onwards, improved’. Overall, the statements with a the idea of the referendum became more large loading on Factor 1 are those, which common. Gradually opinions on the stress the importance of S21 and urge its Nagel and Satoh 1695

Table 3. The results of factor analysis.

FC1 FC2

Keep the construction plan 0.88 20.33 More connected cities and Europe with S21 0.85 0.13 Public communication should be improved 0.73 0.33 Result of stress test is valid 0.67 –0.46 Federal State should finance more 0.67 0.20 Social peace is important 0.59 0.49 Should take account of the other projects 0.57 0.22 State should finance more 0.57 0.13 In the future no major project, if S21 not realised 0.43 –0.32 More public transportation use promoted through S21 0.41 –0.08 Cost calculation and project detail of S21 unclear and doubtful –0.50 0.69 DB should take more cost –0.50 –0.16 Compensation for the cancelled resources –0.66 –0.25 No advantage for the urban development with S21 –0.71 0.29 K21 alternative proposal –0.80 0.37 The building of the S21 will/should be delayed –0.82 –0.26 Not enough profit expected with S21 –0.92 0.24 No further financial agreement –0.94 –0.11 Quorum of plebiscite too high –0.26 0.91 Arbitration should be done 0.43 0.85 More costs will be expected with S21 –0.37 0.82 Referendum should be done –0.11 0.85 Alternative construction plan should be realised 0.27 0.58 Protest should be more controlled 0.27 0.46 City should finance more –0.08 –0.42 Despite the S21, the ruling party can win the election 0.41 –0.55 Container village will be built –0.38 –0.88 S21 should be stopped to save old trees –0.32 0.33 Opponents successful and enough reason –0.19 0.21 Campaign for S21 0.16 0.04 Anger among the citizens against S21 increases –0.13 0.21 EU should take more cost 0.32 –0.30 Involved should take the cost 0.07 –0.25 Other cities should finance more 0.01 –0.01

Loadings 9.95 6.99 Proportion of variation 0.29 0.21

Correlation between factors –0.14

Note: Factors were calculated by minimum residual solution with oblimin rotation by R package psych v.1.6.12 (Revelle, 2016). construction. Therefore, we name Factor 1 Factor 2 consists of the statements calling as the ‘Progress of S21’ factor. Organisations for a referendum (‘Referendum should be that have a large positive score in Factor 1 done’) or are about the ways in which argue for the construction of S21, while the referendum should be implemented organisations that have negative value argue (‘Quorum of plebiscite too high’). for the halt of the construction of S21. Accordingly, we consider this factor as 1696 Urban Studies 56(8)

‘Implementation of Referendum’. Those democracy and the establishment’ (Vento, organisations having positive values on this 2017: 68). Literature in the field of urban score argued for the need of the governance and politics has recently warned Referendum. Quite interestingly, the meta- of a post-democratic or post-political con- category ‘More costs will be expected’ also sensus (MacLeod, 2011), but scholars also belongs to this factor, suggesting that the criticise such theorisations, calling them a actors who argued about the referendum ‘post-political trap’ (Beveridge and Koch, also saw the rise in costs as particularly pro- 2017). While the post-political perspective blematic. Citizens who were against the proj- addresses important developments in urban ect because of the cost explosion called for a politics, its explanatory power is, in our referendum, in order to determine the legiti- view, limited, especially with regards to macy (or lack thereof) of the project. megaprojects. The democratic deficits It should be also noted that the meta- observed by post-democratic scholars (cf. category ‘Protest should be more controlled’ Crouch, 2004) charge that societies have also belongs to this factor with plus (+) load- become less democratic, but they fail to ing. This suggests that some actors arguing account for the multiple modes in which over the referendum also saw the heated pro- ‘democracy’ and democratic participation tests as problematic. and legitimation may be made manifest or The correlation between Factor 1 expressed. We believe that cognitive and (‘Progress of S21’) and Factor 2 ideational perspectives especially (cf. (‘Implementation of Referendum’) was rela- Bradford, 2016) should be examined for a tively low (r = 20.14). This suggests that better understanding of collaborative urban the discussions about the construction of governance and its multiple modes of actua- S21 and the implementation of the lisation, and this is what our paper attempts Referendum belonged to different dimen- to illuminate through its empirical investiga- sions in the discussion. tion of an urban conflict case study using This is a very suggestive result for under- discourse network analysis. standing why the increasing intensity of the To give a more nuanced understanding, protests against S21, and the support for S21 we begin by presenting the history and nar- at the referendum, occurred simultaneously. ratives of the case study Stuttgart 21, which Some actors discussed S21 from the perspec- is an iconic megaproject in the city of tive of the construction and its advantages, Stuttgart, Germany. The protests against in other words, from the perspective of this infrastructural project, the violent esca- growth, as shown in Factor 1. Other actors lation between protestors and the police, the questioned the legitimacy of the plans, or fierce opposition between the conflicting viewed them from the perspective of democ- coalitions, appear to fulfil the diagnoses of racy. These two dimensions (growth and theories of the ‘post-political’: they represent democracy) widened the range of people the citizens’ rejection of the political estab- who consider S21 as problematic, and the lishment, their strong discontent with politi- discussions therefore expanded. cal decision processes, and demands for participation. Discussion and conclusion Our study traced the evolution of the key actors and coalitions in the discourse, and Megaprojects and their focus on expertise our factor analysis identified the most and technocracy, as well as populist opi- important factors or dimensions of the con- nions, play a crucial role in the ‘erosion of flict: on the one hand, the perspective of Nagel and Satoh 1697 growth, and on the other, a perceived lack political structures. In doing so, we compli- of legitimacy of the project. These two cate and qualify the arguments for a ‘post- dimensions, growth and democratic legiti- political’ state of urban politics. Finally, macy, polarised the discourse. future research should continue on this path Through its dynamic and multidimensional of analysing the sources of strong discontent analysis of individual motivation and actor and polarisation, and investigate which issues constellation, our case study reveals that the specifically are the locus of disagreement. motives behind the protests against S21 involved strategic considerations, rather than Acknowledgements the general refusal of the political process that ‘post-political’ theories have suggested. We would like to thank Dorothy Kwek and Jennifer Spohrs for proofreading, and Christin As our study of evolving discourse coali- Eckerle and Prema Hartkopf for assistance. tions shows, there was no single way of addressing the perceived lack of legitimacy in S21 plans. Instead, several different ways Declaration of conflicting interests of tackling this issue emerged over time, The author(s) declared no potential conflicts of including electoral and representative poli- interest with respect to the research, authorship, tics. Our analysis shows that the Green and/or publication of this article. Party, which played an important role in connecting the opponents to S21 and their Funding key issues, and which successfully pushed The author(s) received no financial support for through a state-wide referendum, nonethe- the research, authorship, and/or publication of less had to continue with the megaproject this article. they opposed, given the unexpected results in favour of S21. This, too, was democracy Notes at work. The mediation process, electoral 1. These were initially collected in a research politics, and referendum show how several project (Nagel, 2014) on the dynamics of the decision-making tools available to the politi- polarised discourse and the identification of cal system were able to defuse the conflict. the fields of conflict, and later published in Our case study shows that the democratic 2016. 2. From www.philipleifeld.com. deficits postulated by theories of the ‘post- 3. Those issues include possible cost hikes and political’ are not valid in the case of S21. environmental damage. Admittedly, it is not advisable to draw 4. www.visone.info. general conclusions from a single case study, 5. http://visone.info/wiki/index.php/ but despite certain limitations and biases Quick_layout. implied by the methodology (e.g. newspaper data, coder-reliability) our study contributes References to recent urban studies literature in several ways. First, our case study of S21 sheds light Banister D and Hall P (1993) The second railway age. Built Environment 19(3/4): 156–162. on the evolution and management of con- Barranco J and Wisler D (1999) Validity and sys- flicts over megaprojects. Second, we show tematicity of newspaper data in event analysis. how research in urban politics and especially European Sociological Review 16(3): 301–322. on megaprojects must look beyond the obvi- Baumgarten B and Grauel J (2009) The theoreti- ous facts and explicit statements, and go cal potential of website and newspaper data for deeper to explore how political issues and analysing political communication processes. concerns combine with relational social and Historical Social Research 34(1): 94–121. 1698 Urban Studies 56(8)

Beveridge R and Koch P (2017) The post-political Gissendanner S (2004) Mayors, governance coali- trap? Reflections on politics, agency and the tions, and strategic capacity. Drawing lessons city. Urban Studies 54(1): 31–43. from Germany for theories of urban govern- Blu¨hdorn I (2013) The governance of unsustain- ance. Urban Affairs Review 40(1): 44–77. ability: Ecology and democracy after the post- Graham S and Marvin S (2001) Splintering democratic turn. Environmental Politics 22(1): Urbanism: Networked Infrastructures, Techno- 16–36. logical Mobilities and the Urban Condition. Boland P (2014) The relationship between spatial Oxford: Blackwell. planning and economic competitiveness: The Graham S and Marvin S (2003) Planning cyber- ‘path to economic nirvana’ or a ‘dangerous cities? Integrating telecommunications into obsession’? Environment and Planning A 46: urban planning. In: Helten F and Fischer B 770–787. (eds) Envisioning Telecity: Towards the Urbani- Bradford N (2016) Ideas and collaborative gov- sation of ICT. New York: Peter Lang Publish- ernance. Urban Affairs Review 52(5): ing, pp. 23–59. S659–684. Hage P and Haray F (1983) Structural Models in Bulkeley H (2000) Discourse coalitions and the Anthropology. Cambridge: Cambridge Univer- Australian Climate Change Policy Network. sity Press. Environment and Planning C: Government and Hajer M (1996) Ecological modernization as cul- Policy 18(6): 727–748. tural politics. In: Lash S, Szerszynski B and Bylund J (2012) Postpolitical correctness? Plan- Wynne B (eds) Risk, Environment and Moder- ning Theory 11(3): 319–327. nity: Towards a New Ecology. London: SAGE Crouch C (2004) Post-democracy. Cambridge: Publications Ltd, pp. 246–268. Polity Press. Hajer M (2002) Discourse analysis and the study Douglass M and Huang L (2007) Globalizing the of policy making. European Political Science city in southeast Asia: Utopia on the urban 2(1): 61–65. edge – The case of Phu My Hung, Saigon. Hajer MA (1993) Discourse coalitions and the International Journal of Asia–Pacific Studies 3: institutionalization of practice: The case of 1–42. acid rain in Britain. In: Fischer F and Forester Earl J, Martin A, McCarthy J, et al. (2004) The J (eds) Discourse Coalitions and the Institutio- use of newspaper data in the study of collec- nalization of Practice: The Case of Acid Rain tive action source. Annual Review of Sociology in Britain, The Argumentative Turn in Policy 30: 65–80. Analysis and Planning. Durham, NC, and Lon- Fischer F (2003) Reframing Public Policy: Discur- don: Duke University Press, pp. 43–76. sive Politics and Deliberative Practices. New Hajer MA (1995) The Politics of Environmental York: Oxford University Press. Discourse: Ecological Modernization and the Fischer F and Forester J (ed) (1993) The Argu- Policy Process. Oxford: Claredon Press. mentative Turn in Policy Analysis and Planning. Hilgartner S and Bosk CL (1988) The rise and fall Durham, NC: Duke University Press. of social problems: A public arenas model. Fisher DR, Leifeld P and Iwaki Y (2013) Map- American Journal of Sociology 94(1): 53–78. ping the ideological networks of American cli- Hornig E-C (2017) Airport expansions and public mate politics. Climatic Change 116(3): protests – The democratic dilemma of verti- 523–545. cally asymmetric policies. European Policy Franzosi R (1987) The press as a source of socio- Analysis 3(2): 324–342. historical data: Issues in the methodology of Iacobucci D (1994) Graphs and matrices. In: Was- data collection from newspapers. Historical sermann S and Faust K (eds) Social Network Methods: A Journal of Quantitative and Inter- Analysis. Methods and Applications. Cambridge: disciplinary History 20(1): 5–16. Cambridge University Press, pp. 92–166. Franzosi R (2004) From Words to Numbers. Nar- Janning F, Leifeld P, Malang T, et al. (2009) Dis- rative, Data, and Social Science. Cambridge: kursnetzwerkanalyse. U¨berlegungen zur Theo- Cambridge University Press. riebildung und Methodik. In: Schneider V, Nagel and Satoh 1699

Janning F, Leifeld P and Malang T (eds) Poli- Joint Sessions of Workshops, Mu¨nster, 22–27 tiknetzwerke. Modelle, Anwendungen und March. Available online at: http://www. Visualisierungen. Wiesbaden: VS Verlag fu¨r philipleifeld.de. Sozialwissenschaften, pp. 59–92. Leifeld P (2012) Discourse Network Analyzer Jenkins-Smith HC, et al. (2014) The advocacy (DNA) manual. Bonn, Germany: Max Planck coalition framework: Foundations, evolution, Institute for Research on Collective Goods. and ongoing research. In: Sabatier PA and Available at: http://www.philipleifeld.de. Weible C (eds) Theories of the Policy Process Leifeld P (2013) Reconceptualizing major policy (3rd edn). Boulder, CO: Westview Press, pp. change in the advocacy coalition framework. 183–223. A discourse network analysis of German pen- Jungherr A and Ju¨rgens P (2014) Through a glass, sion politics. The Policy Studies Journal 41(1): darkly: Tactical support and symbolic associa- 169–198. tion in Twitter messages commenting on Stutt- Leifeld P and Haunss S (2012) Political discourse gart 21. Social Science Computer Review 32(1): networks and the conflict over software 74–89. patents in Europe. European Journal of Politi- Kenis P and Schneider V (1991) Policy networks cal Research 51: 382–409. and policy analysis: Scrutinizing a new analyti- McCann E (2017) Governing urbanism. Urban cal toolbox. In: Marin B and Mayntz R (eds) governance studies 1.0, 2.0 and beyond. Urban Policy Networks: Empirical Evidence and The- Studies 54(2): S312–326. oretical Considerations. Frankfurt am Main: McCarthy JD, McPhail C and Smith J (1996) Campus, pp. 25–59. Images of protest: Dimensions of selection Koopmans R (1996) Explaining the rise of racist bias in media coverage of Washington demon- and extreme right violence in Western Europe: strations, 1982 to 1991. American Sociological Grievances or opportunities? European Journal Review 61(3): 478–499. of Political Research 30(2): 185–216. McCombs M and Shaw D (1972) The agenda- Koopmans R and Rucht D (1999) Protest event setting function of mass media. Public Opinion analysis – Where to now? Mobilization 4: Quarterly 36: 176–187. 123–130. MacLeod G (2011) Urban politics reconsidered: Koopmans R and Statham P (1999) Political Growth machine to post-democratic city? claims analysis: Integrating protest event and Urban Studies 48(12): 2629–2660. political discourse approaches. Mobilization Maier T (2013) ‘Stuttgart 21’ im Spiegel von 4(2): 203–221. Facebook-Aktivita¨ten der Befu¨rworter und Kriesi H (ed.) (1995) New Social Movements in der Gegner des Projektes. In: Brettschneider F Western Europe: A Comparative Analysis. Vol- and Schuster W (eds) Stuttgart 21. Ein Groß- ume 5. Minneapolis, MN: University of Min- projekt zwischen Protest und Akzeptanz. Wies- nesota Press. baden: Springer Fachmedien, pp. 77–96. Legacy C (2016) Transforming transport planning Marres N (2005) No issue, no public: Democratic in the postpolitical era. Urban Studies 53(14): deficits after the displacement of politics.PhD S3108–3124. thesis, University of Amsterdam, The Nether- Leifeld P (2009) Die Untersuchung von Diskurs- lands [Cited from Legacy (2016)]. netzwerken mit dem Discourse Network Ana- Mouffe C (2005) On the Political. London: lyzer (DNA). In: Schneider V, et al. (eds) Routledge. Politiknetzwerke. Modelle, Anwendungen und Moulaert F, Rodriguez A and Swyngedouw E Visualisierungen. Wiesbaden: VS Verlag, pp. (eds) (2003) The Globalised City: Economic 391–404. Restructuring and Social Polarization in Eur- Leifeld P (2010) Political discourse networks: The opean Cities. Oxford: Oxford University Press. missing link in the study of policy oriented dis- Moulaert F, Swyngedouw E and Rodriguez A course. Paper presented at the workshop (2001) Social polarization in metropolitan ‘Ideas, Policy Design and Policy Instruments: areas. European Urban and Regional Studies 8: Casting Light on the Missing Link’, ECPR 99–102. 1700 Urban Studies 56(8)

Muller A (2015) Using discourse network analysis in political engagement. Social Justice to measure discourse coalitions: Towards a Research 27(1): 24–44. formal analysis of political discourse. World Rucht D, Koopmans R, Niedhardt F, et al. Political Science 11(2): 377–440. (1999) Acts of Dissent: New Developments in Nagel M (2014) Polarisierung im politischen Dis- the Study of Protest. Oxford: Rowman & Lit- kurs. Eine Analyse dynamischer Wissen- und tlefield Publishers. Diskursnetzwerke am Beispiel von ‘Stuttgart’. Salet W (2008) Rethinking urban projects. PhD Dissertation, University of Konstanz. Experiences in Europe. Urban Studies 45(11): Nagel M (2016) Polarisierung im politischen Dis- 2343–2363. kurs. Eine Netzwerkanalyse zum Konflikt um Schmidt VA (2008) Discursive institutionalism: ‘Stuttgart’. Wiesbaden: Springer. The explanatory power of ideas and discourse. Novy J and Peters D (2013) Railway megapro- Annual Review of Political Science 11: 303– jects as catalysts for the re-making of post- 326. industrial cities? The case of Stuttgart 21 in Schmidt VA and Radaelli CM (2004) Policy Germany. In: Del Cerro Santamarı´a G (ed) change and discourse in Europe: Conceptual Urban Megaprojects: A Worldwide View. and methodological issues. West European Research in Urban Sociology 13. Bingley: Politics 27(2): 183–210. Emerald Group Publishing, pp. 237–262. Swyngedouw E, Moulaert F and Rodriguez A Oosterlynck S and Swyngedouw E (2010) Noise (2002) Neoliberal urbanization in Europe: reduction: The postpolitical quandary of night Large-scale urban development projects and flights at Brussels airport. Environment and the new urban policy. Antipode 34: 542–577. Planning A 42: 1577–1594. Szarka J (2004) Wind power, discourse coalitions Ortiz DG, Myers DD, Walls NE, et al. (2005) and climate change: Breaking the stalemate? Where do we stand with the newspaper data? European Environment 14(6): 314–330. Mobilization: An International Journal 10: Vatter A and Heidelberger A (2013) Volks- 397–419. entscheide nach dem NIMBY-Prinzip? Eine Rancie` re J (2001) Ten theses on politics. Theory Analyse des Abstimmungsverhaltens zu Stutt- and Event 5(3): 17–34. gart 21. (About the voting behavior and Rancie` re J (2009) Hatred of Democracy. London: NIMBY effect). Politische Vierteljahresschrift Verso. 54(2): 317–336. Revelle W (2016) psych: Procedures for Personality Vento AT (2017) Mega-project meltdown: Post- and Psychological Research, Northwestern Uni- politics, neoliberal urban regeneration and versity, Evanston, Illinois, USA. Available at: Valencia’s fiscal crisis. Urban Studies 54(1): https://CRAN.R-project.org/package=psych 68–84. (Version = 1.6.12). Young D and Keil R (2010) Reconnecting the Rothmund T, Baumert A and Zinkernagel A disconnected: The politics of infrastructure in (2014) The German ‘Wutbu¨rger’: How justice the in-between city. Cities 27: 87–95. sensitivity accounts for individual differences